This week marks year two of attending the Well Site Automation conference in Houston. Last year there were lots of presenters who mentioned MQTT and IoT architectures. This year the focus was on data analytics, but IoT came up lots in the presentations.
I only attended two days this year. Day one definitely had a more commercial feel with a strong sales presentation by eLynx for their data analytics service. The other presenters were pretty generic but had some interesting things they were doing.
We heard from Devon Energy and their water processing facility. They are planning to bring all the data back using MQTT, an IoT protocol. That was followed up by a presentation from Encana who explained how they are gaining operational efficiency through the use of an Operations Control Center (or decision center). Encana also stressed the need for management buy-in for automation projects because ROI is sometimes very difficult to calculate. Their stated goal was to maintain a flat head count even though they are adding 60-100 wells per year for the next 5 years.
Whiting Petroleum made a presentation about collecting data for analytics. They claimed to be collecting 5-30 second data on 80,000 points. I did some math on that and its up in the 30-40GB/day range before communication protocol overhead. So I doubt they are keeping all that data, but sure sounded impressive. They tried a cloud historian service and found that it could not keep up (no way!) and reverted back to a traditional historian vendor (OSIsoft Pi?). Whiting recommended putting context around the data as close to the field device as possible, and they were on a mission to add equipment tags and other information into their PLC and RTU code to help maintain that context and data integrity.
The eLynx presentation was next. Not much to report here, other than a very strong partnership with Microsoft and the fact they indicated that their data analytics models are not quite ready for prime time. You could tell a lot of the people in the room tuned this presentation out. This conference is more about sharing what people are doing for advancing well site automation, not sharing your sales pitch.
Day two was more about analytics. Devon Energy presented again. Their data scientists had a great presentation with some valid use cases, including a failed one. They were looking at analysing drilling rig downtime and prediciting ECP (submersible pump) failure rates. The drilling rig downtime analysis was successful. They used logged downtime and notes from each day on the rig to identify problem contractors, repeat equipment failures and more. The submersible pump failure analysis was not successful because the data used to create the model was not very clean (failures were denoted differently in different areas) so the model worked well on the training data but when analyzing new data.
Unfortunately I had to skip out early on Day Two to get back up to where I was staying for the next couple nights. But I did manage to make some new connections in the US and I am really excited about what is in store for 2018.